This is a multicenter, randomized double blind trial of Zidovudine (ZDV) versus 2',3' dideoxyinosine (ddI) versus a combination of ZDV and ddI in symptomatic HIV-infected children to evaluate the safety, tolerance, and efficacy of the treatment arms. All drugs were administered orally. Evaluation of response was based on survival, disease progression, development of opportunistic infections, growth parameters, and neuropsychological function. Under this protocol, neither the physicians nor the patients knew which therapy each patient received. However, the monitoring board's review revealed that the group of children receiving ZDV alone had more rapid rates of disease progression as measured by growth failure, new opportunistic infections, neurologic and neurodevelopmental deterioration or death. The major side effects seen in the study were blood and other chemical abnormalities. Children receiving ZDV monotherapy had a significantly higher proportion of these side effects. In February 1995, the study continued without the single agent ZDV arm. Because differences in disease progression and side effects between the other two arms were not significant, the ddI and ddI/ZDV combination arms continued. Children who received ZDV alone were offered the option of switching to another therapy or continuing on ZDV, whichever was appropriate to the individual child's condition. Further analysis of the study, including determinations of any differences in effectiveness and toxicity between treatment with ddI and with a combination of ddI and ZDV, is currently pending at this time. The study officially concluded in August, 1995.